[prev in list] [next in list] [prev in thread] [next in thread] 

List:       xfree-newbie
Subject:    Re: [Newbie]Re: G400
From:       dep <dennispowell () earthlink ! net>
Date:       2001-06-26 8:38:37
[Download RAW message or body]

On Monday 25 June 2001 11:29 pm, Derek L Davies wrote:

| I don't run suse and don't have the 3Ddiag program you mentioned. 
| Can I get 3Ddiag running on debian?  Hmm, I keep hearing such nice
| things about suse...

i doubt 3Ddiag would work on debian -- too many differences. and 
yeah, suse is nice, from an ease-of-use point of view, though i don't 
think anyone wouyld argue that it's more solid or otherwise better 
than a well-tuned and updated debian box.

| Still running through various setups with the kernel.  So far,
| agpgart and mga manually loaded as modules prior to startx yeilds
| the most "promising" log (ie, DRM seems more chatty without any
| indication of error), but still no real speedup.

my advice was based on a lot of experience in trying things out. for 
instance, for a period of time, from the linux-2.4.0-pre series and 
xfree86-4.01 until just this week, when i switched to 2.4.5 and 
xfree-4.03, i could not get direct-rendering to work properly without 
actually taking the appropriate stuff from the xfree source tree and 
using it to replace some files of the same name in the kernel source 
tree, and then build the kernel. this week i discovered that that no 
longer works, but compiling the drivers into the kernel, using the 
code supplied in the kernel source, does. in short, the whole thing 
is as flaky as all get out, and figuring what will work has at least 
here been a matter of lots of trial, lots of error, but ultimate 
(though inexplicable) success.

| Is it known that the module approch *doesn't* work?  I don't really
| care either way except that I seem to remember that it was
| important to load them as modules from the experience I had getting
| it working originally.  I seem to recall that there was some caveat
| or trick related to kernel config.

well, it was once true that it was far less likely to work if 
framebuffer support was enabled. as to the module approach -- it 
certainly doesn't work here, but that's not definitive. 
(additionally, i do not know what if any effect the fact that i'm 
running, and compiling kernels for, an AMD chip has.)
-- 
dep
 
there's more to history than what's in books;
that's why it took so long to happen.
_______________________________________________
Newbie mailing list
Newbie@XFree86.Org
*** To unsubscribe , or change message options, see:
http://XFree86.Org/mailman/listinfo/newbie

[prev in list] [next in list] [prev in thread] [next in thread]